Chiral workpiece attitude control method supporting multi-scene application

Abstract

The invention relates to the technical field of chiral workpiece attitude control, and discloses a chiral workpiece attitude control method supporting multi-scene application, which comprises the following steps: receiving a CAD model of a new workpiece, analyzing geometric characteristics, determining chiral symmetry information, and generating a corresponding mirror image model as a chiral template; constructing parameter templates of different workpiece types, determining the workpiece types of the chiral template, and performing dynamic adaptation in combination with the parameter templates to obtain parameter configuration information; a plurality of candidate postures of the chiral template are generated, key point locations are marked according to task requirements, posture configuration rules are generated, the candidate postures are screened, and the optimal posture is determined; according to the method, the workpiece attitude data is obtained, whether the attitude adjustment instruction is generated or not is judged by combining the optimal attitude, the dual-channel feature extraction result and the parameter configuration information, the problems that in the prior art, model processing precision is insufficient, parameter adaptability is poor, the optimal attitude is difficult to determine, adjustment is not timely and the like are solved, and the stability and efficiency of the manufacturing link are improved.

Description

Technical Field

[0001] This application relates to the field of chiral workpiece posture control technology, and in particular to a chiral workpiece posture control method that supports multiple application scenarios. Background Technology

[0002] The automated assembly, intelligent inspection, and robotic grasping of workpieces all rely on the precise processing of workpiece models and the efficient control of their posture. As product complexity increases, workpieces with chiral symmetry characteristics are being used more and more widely. These workpieces place stringent requirements on the accuracy of model processing and the stability of posture control.

[0003] In the existing technology, there is a lack of model processing methods for chiral workpieces for different application scenarios, and the parameter configuration adaptability is insufficient, making it difficult to meet the needs of the scenario and determine the optimal posture. At the same time, there is a lack of closed-loop management logic after the posture control command is issued, which leads to unreasonable posture and untimely adjustment, which can easily cause assembly failure or workpiece damage, thereby reducing the stability and efficiency of the manufacturing process. Summary of the Invention

[0004] To address the aforementioned technical issues, this application provides a chiral workpiece posture control method supporting multi-scenario applications. By generating a chiral template of a new workpiece and determining the workpiece category, and combining it with a pre-built parameter template to determine parameter configuration information, several candidate postures are generated. Based on key points and posture configuration rules, the optimal posture is determined. Workpiece posture data is acquired in real time, and it is determined whether posture adjustment is required. This method effectively solves the problems of insufficient model processing accuracy, poor parameter adaptability, difficulty in determining the optimal posture, and untimely adjustment in the prior art, significantly improving the stability and efficiency of the manufacturing process.

[0005] In some embodiments of this application, a chiral workpiece posture control method supporting multi-scenario applications is provided, including: Receive the CAD model of the new workpiece and analyze its geometric features. Determine the chiral symmetry information based on the analysis results and generate the corresponding mirror model as a chiral template. Parameter templates for different workpiece categories are pre-built, the workpiece category of the chiral template is determined, and dynamic adaptation is performed in combination with the corresponding parameter template to obtain the parameter configuration information of the chiral template. Generate several candidate poses for the chiral template, mark key points according to task requirements and generate pose configuration rules, filter several candidate poses according to key points and pose configuration rules, determine the optimal pose and issue pose control commands. The system acquires workpiece posture data in real time, and combines the optimal posture, dual-channel feature extraction results, and parameter configuration information to determine whether to generate a posture adjustment command.

[0006] In some embodiments of this application, receiving a CAD model of a new workpiece and analyzing its geometric features, determining chiral symmetry information based on the analysis results, and generating a corresponding mirror model as a chiral template includes: The CAD model of the new workpiece is received and standardized, including coordinate normalization, mesh simplification and noise removal. Extract the geometric features of the processed CAD model, including curvature features, normal vector features, feature lines, and topological features; Analyze the geometric features and determine the candidate axes of symmetry and candidate planes of symmetry based on the analysis results; Calculate the degree of symmetry of candidate axes of symmetry and candidate planes of symmetry; Candidate symmetry axes and candidate symmetry planes with a symmetry degree greater than a preset symmetry degree threshold are set as chiral symmetry information, and a mirror model is generated. Calculate the matching degree between the mirror model and the CAD model; If the matching degree is greater than the preset matching degree threshold, the mirror model is set as a chiral template. If the matching degree is not greater than the preset matching degree threshold, the mirror model is corrected until the matching degree is greater than the preset matching degree threshold, and the corresponding mirror model is set as a chiral template.

[0007] In some embodiments of this application, parameter templates for different workpiece categories are pre-built, including: Feature extraction is performed on historical workpiece data to obtain several workpiece features; Multiple workpiece categories are generated based on all workpiece features, and a parameter feature reference library is constructed for each workpiece category; Several demand evaluation indicators are pre-defined; The parameter feature reference library for each workpiece category is evaluated based on several demand evaluation indicators to obtain the demand evaluation value for each demand evaluation indicator. The demand evaluation indicators for the same workpiece category are sorted according to the demand evaluation value to obtain a demand evaluation indicator sequence; Based on the ranking results in the demand evaluation index sequence and the corresponding demand evaluation values, key parameters and corresponding parameter value ranges are set for each workpiece category. A parameter template for each workpiece category is constructed based on the key parameters and corresponding parameter value ranges for each workpiece category.

[0008] In some embodiments of this application, the workpiece category of the chiral template is determined and dynamically adapted in conjunction with the corresponding parameter template to obtain the parameter configuration information of the chiral template, including: Extract real-time workpiece features from chiral templates; The real-time workpiece features are compared with several workpiece features of each workpiece category to obtain several similarities between the chiral template and different workpiece categories. The formula for calculating the similarity is: ; Where D represents the similarity, and n1 represents the number of feature evaluation indicators. t2i is the reference evaluation value of the i-th feature evaluation index generated for real-time workpiece features, t2i is the historical evaluation value of the i-th feature evaluation index generated for workpiece features, ai is the weight coefficient of the i-th feature evaluation index, and n2 is the number of demand evaluation indicators. The reference demand evaluation value for the s-th demand evaluation index generated for real-time workpiece features. The demand evaluation value of the s-th demand evaluation index generated for the workpiece features, gs is the weight coefficient of the s-th demand evaluation index, d1 is the first similarity conversion coefficient, and d2 is the second similarity conversion coefficient; All workpiece categories are sorted according to similarity, and the workpiece category with the highest similarity is set as the workpiece category of the chiral template. Based on the workpiece category and the size information of the new workpiece, the parameter template of the corresponding workpiece category is loaded and dynamically adjusted to generate the parameter configuration information of the chiral template.

[0009] In some embodiments of this application, several candidate poses for generating a chiral template are included: Several first candidate poses of the chiral template are generated based on preset coarse-grained sampling parameters; For each first candidate pose, contour analysis is performed to extract contour features, including contour shape, contour size, and contour curvature. Compare the degree of difference in contour features of different first candidate poses, sort the contour features according to the degree of difference, and set several feature matching points for the corresponding contour features based on the sorting results. Each first candidate pose is projected onto the 3D point cloud image of the new workpiece, and the matching error between each contour feature of the first candidate pose and several feature matching points of the corresponding contour feature in the 3D point cloud image is calculated. The comprehensive matching error of the corresponding contour feature is generated based on the matching error of all feature matching points of each contour feature, and the effective coefficient of the corresponding first candidate pose is generated by combining the weight coefficient of the corresponding contour feature. Eliminate the first candidate pose whose effective coefficient is less than the preset first effective coefficient, and determine the effective region based on the remaining first candidate poses; Fine-grained sampling is performed on the effective region, and the candidate poses of the chiral template are determined based on the sampling results and the remaining first candidate poses.

[0010] In some embodiments of this application, fine-grained sampling is performed on the effective region, and candidate poses of the chiral template are determined based on the sampling results and the remaining first candidate poses, including: Based on the functional characteristics and task requirements of the new workpiece, the feature direction is determined. It is then determined whether the effective area covers the feature direction. If not, several first candidate postures are generated around the feature direction and the effective coefficient is calculated until the effective area covers the feature direction. If so, the effective area is sampled based on the preset fine-grained sampling parameters to obtain several second candidate postures. A multi-dimensional feature matching is performed on several second candidate poses, and an effective coefficient for each second candidate pose is generated based on the matching results. The multi-dimensional features include local point features, texture features, and topological structure features. Eliminate second candidate postures whose effective coefficient is less than the preset second effective coefficient; The remaining first candidate poses and the remaining second candidate poses are compared. Duplicate poses are eliminated based on the comparison results. The remaining first candidate poses and the remaining second candidate poses are set as the final candidate poses of the chiral template.

[0011] In some embodiments of this application, several candidate poses are screened based on key points and pose configuration rules to determine the optimal pose, including: Determine the task requirements for the new workpiece; Based on task requirements and user interaction information, graphical annotation and language rule conversion are performed to obtain several key points and several posture configuration rules. Calculate the correlation coefficient between each attitude configuration rule and several key points. Set the key points with correlation coefficients greater than the preset correlation coefficient threshold as the associated key points of the attitude configuration rule, and generate the standard constraint conditions for each associated key point to satisfy the attitude configuration rule. The associated key point sequence for each attitude configuration rule is generated sequentially. The order of the associated key point sequence is set according to the association coefficient, and each associated key point is mapped to a corresponding weight coefficient and standard constraint conditions. Each pose matching rule is treated as an optimization objective, and the weight coefficient of each optimization objective is set according to the task requirements; For each candidate pose, perform key point matching analysis to determine whether the candidate pose satisfies the standard constraint conditions of each associated key point in the associated key point sequence of each optimization objective. Based on the judgment results, determine the first number of associated key points that satisfy the standard constraint conditions of the same optimization objective and the second number of associated key points that do not satisfy the standard constraint conditions of the same optimization objective. The optimization coefficient of the corresponding candidate pose is calculated based on the first and second quantities of the same candidate pose for all optimization objectives. Several candidate poses are sorted according to the optimization coefficient, and the candidate pose ranked first is set as the optimal pose.

[0012] In some embodiments of this application, the optimization coefficient of the corresponding candidate pose is calculated based on a first number and a second number of the same candidate pose for all optimization objectives, including: The formula for calculating the preference coefficient is: ; Where Y is the optimization coefficient, and w is the number of optimization targets. This represents the first number of associated key points that satisfy the v-th optimization objective for the corresponding candidate pose. Let y1 be the second number of associated key points corresponding to candidate poses that do not meet the v-th optimization objective, and y2 be the first preferred transformation coefficient and y2 be the second preferred transformation coefficient. The weight coefficient of the c-th associated key point in the v-th optimization objective is determined by... is the weight coefficient for the v-th optimization objective.

[0013] In some embodiments of this application, workpiece posture data is acquired in real time, and combined with the optimal posture, dual-channel feature extraction results, and parameter configuration information, it is determined whether to generate a posture control command, including: Determine the real-time attitude based on the workpiece attitude data; Calculate the degree of positional deviation, angular deviation, and motion trend deviation between the real-time attitude and the optimal attitude, and quantify them into a first deviation coefficient, a second deviation coefficient, and a third deviation coefficient; Based on the dual-channel feature extraction results, the geometric feature deviation and chiral feature deviation of the workpiece surface are obtained respectively, and quantified into the fourth deviation coefficient and the fifth deviation coefficient. Determine the parameter value range of the chiral template in the parameter configuration information, and determine whether the real-time attitude exceeds the parameter value range. If so, quantize to obtain the sixth deviation coefficient. The above deviation coefficients and their corresponding weighting coefficients are weighted and summed to obtain the comprehensive deviation coefficient. Whether to generate attitude control commands is determined based on the comprehensive deviation coefficient.

[0014] In some embodiments of this application, determining whether to generate an attitude control command based on a comprehensive deviation coefficient includes: Pre-set the overall deviation coefficient threshold; If the overall deviation coefficient is less than the overall deviation coefficient threshold, an attitude control command is generated. If the overall deviation coefficient is not less than the overall deviation coefficient threshold, no attitude control command will be generated.

[0015] The chiral workpiece posture control method supporting multiple application scenarios according to this application has the following advantages compared with the prior art: By generating a chiral template of a new workpiece and determining the workpiece category, and combining it with a pre-built parameter template to determine parameter configuration information, several candidate postures are generated. Based on key points and posture configuration rules, the optimal posture is determined. Workpiece posture data is acquired in real time and it is determined whether posture adjustment is required. This effectively solves the problems of insufficient model processing accuracy, poor parameter adaptability, difficulty in determining the optimal posture, and untimely adjustment in the existing technology, and significantly improves the stability and efficiency of the manufacturing process. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating a chiral workpiece posture control method supporting multiple application scenarios in an embodiment of this application. Detailed Implementation

[0017] The specific embodiments of this application will be described in further detail below with reference to the accompanying drawings and examples. The following examples are used to illustrate this application, but are not intended to limit the scope of this application.

[0018] In the description of this application, it should be understood that the terms "center", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application.

[0019] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0020] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection between two components. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0021] like Figure 1As shown in the figure, an embodiment of this application provides a chiral workpiece posture control method supporting multiple application scenarios, comprising: S101: Receive the CAD model of the new workpiece and analyze its geometric features. Based on the analysis results, determine the chiral symmetry information and generate the corresponding mirror model as a chiral template. S102: Pre-build parameter templates for different workpiece categories, determine the workpiece category of the chiral template, and dynamically adapt it in combination with the corresponding parameter template to obtain the parameter configuration information of the chiral template. S103: Generate several candidate poses for the chiral template, mark key points according to task requirements and generate pose configuration rules, filter several candidate poses according to key points and pose configuration rules, determine the optimal pose and issue pose control commands. S104: Acquire workpiece posture data in real time, and combine the optimal posture, dual-channel feature extraction results, and parameter configuration information to determine whether to generate a posture adjustment command.

[0022] In some embodiments of this application, receiving a CAD model of a new workpiece and analyzing its geometric features, determining chiral symmetry information based on the analysis results, and generating a corresponding mirror model as a chiral template includes: The CAD model of the new workpiece is received and standardized, including coordinate normalization, mesh simplification and noise removal. Extract the geometric features of the processed CAD model, including curvature features, normal vector features, feature lines, and topological features; Analyze the geometric features and determine the candidate axes of symmetry and candidate planes of symmetry based on the analysis results; Calculate the degree of symmetry of candidate axes of symmetry and candidate planes of symmetry; Candidate symmetry axes and candidate symmetry planes with a symmetry degree greater than a preset symmetry degree threshold are set as chiral symmetry information, and a mirror model is generated. Calculate the matching degree between the mirror model and the CAD model; If the matching degree is greater than the preset matching degree threshold, the mirror model is set as a chiral template. If the matching degree is not greater than the preset matching degree threshold, the mirror model is corrected until the matching degree is greater than the preset matching degree threshold, and the corresponding mirror model is set as a chiral template.

[0023] In this embodiment, coordinate normalization refers to translating the CAD model to a coordinate system with the centroid as the origin and scaling it to a unit bounding box. Mesh simplification refers to simplifying the high-density mesh, retaining key geometric features while reducing computational complexity. Noise removal refers to using a bilateral filtering algorithm to smooth surface noise and retain sharp features.

[0024] In this embodiment, curvature features refer to calculating the Gaussian curvature and average curvature of each vertex to identify sharp edges and smooth regions. Normal vector features refer to calculating the normal vector of each vertex to analyze the surface orientation distribution. Feature lines refer to key feature lines such as edges and ridges based on the curvature abrupt change detection model. Topological structure features refer to identifying features such as holes, protrusions, and depressions in the model. By analyzing the model surface and extracting geometric features, and classifying regions with similar geometric features, regions with obvious symmetry attributes are identified, laying the foundation for subsequent construction of chiral templates and improving accuracy.

[0025] In this embodiment, candidate symmetry axes and candidate symmetry planes refer to axes and planes that have symmetry relationships, and the preset symmetry threshold is the minimum degree of symmetry required to measure whether a candidate symmetry axis / plane can serve as a chiral reference.

[0026] In this embodiment, the mirror model is a model obtained by performing a "spatial mirror transformation" on the original CAD model based on the filtered chiral symmetry information (symmetry axis / plane). Its generation logic is to keep the geometric features and topological relationships of the model unchanged, and only reverse the spatial orientation.

[0027] In some embodiments of this application, parameter templates for different workpiece categories are pre-built, including: Feature extraction is performed on historical workpiece data to obtain several workpiece features; Multiple workpiece categories are generated based on all workpiece features, and a parameter feature reference library is constructed for each workpiece category; Several demand evaluation indicators are pre-defined; The parameter feature reference library for each workpiece category is evaluated based on several demand evaluation indicators to obtain the demand evaluation value for each demand evaluation indicator. The demand evaluation indicators for the same workpiece category are sorted according to the demand evaluation value to obtain a demand evaluation indicator sequence; Based on the ranking results in the demand evaluation index sequence and the corresponding demand evaluation values, key parameters and corresponding parameter value ranges are set for each workpiece category. A parameter template for each workpiece category is constructed based on the key parameters and corresponding parameter value ranges for each workpiece category.

[0028] In this embodiment, the workpiece features include, but are not limited to, the physical characteristics of the workpiece itself (size and precision, appearance material and chiral characteristics), the requirements for the vision system (sensor selection and recognition speed), the final goal and decision logic.

[0029] In this embodiment, the requirements evaluation indicators include workpiece characteristics, accuracy requirements, task objectives, and environmental interference requirements.

[0030] In this embodiment, based on the demand evaluation indicators ranked first in the demand evaluation indicator sequence and their corresponding demand evaluation values, the key parameters and their value ranges in the workpiece category parameter template are determined. For demand evaluation indicators ranked lower, it can be decided whether to include them in the parameter template or to give them a lower weight, depending on the actual situation. At the same time, considering the commonalities and differences between different workpiece categories, common parameters can use the same value range or calculation method in multiple workpiece category parameter templates, while different parameters are set separately according to the characteristics of each workpiece category. In addition, as new workpiece data is continuously accumulated and process requirements are updated, the parameter template is regularly evaluated and optimized, outdated or unreasonable parameters are deleted, and new key parameters are added to ensure that the parameter template always adapts to the needs of multi-scenario applications.

[0031] In this embodiment, the parameter template includes workpiece material parameters, machining accuracy parameters, assembly adaptation parameters, and feature weight parameters for the same workpiece category.

[0032] In some embodiments of this application, the workpiece category of the chiral template is determined and dynamically adapted in conjunction with the corresponding parameter template to obtain the parameter configuration information of the chiral template, including: Extract real-time workpiece features from chiral templates; The real-time workpiece features are compared with several workpiece features of each workpiece category to obtain several similarities between the chiral template and different workpiece categories. The formula for calculating the similarity is: ; Where D represents the similarity, and n1 represents the number of feature evaluation indicators. t2i is the reference evaluation value of the i-th feature evaluation index generated for real-time workpiece features, t2i is the historical evaluation value of the i-th feature evaluation index generated for workpiece features, ai is the weight coefficient of the i-th feature evaluation index, and n2 is the number of demand evaluation indicators. The reference demand evaluation value for the s-th demand evaluation index generated for real-time workpiece features. The demand evaluation value of the s-th demand evaluation index generated for the workpiece features, gs is the weight coefficient of the s-th demand evaluation index, d1 is the first similarity conversion coefficient, and d2 is the second similarity conversion coefficient; All workpiece categories are sorted according to similarity, and the workpiece category with the highest similarity is set as the workpiece category of the chiral template. Based on the workpiece category and the size information of the new workpiece, the parameter template of the corresponding workpiece category is loaded and dynamically adjusted to generate the parameter configuration information of the chiral template.

[0033] In this embodiment, the real-time workpiece features cover multiple dimensions such as geometric size features, surface texture features, and chiral structure features. The accuracy of feature matching is ensured through comprehensive comparison.

[0034] In this embodiment, the feature evaluation indicators are set based on the real-time workpiece characteristics of the chiral template, including geometric shape matching degree, dimensional tolerance compliance degree, and surface quality assessment value. The feature evaluation indicators and the requirement evaluation indicators together constitute a multi-dimensional standard for evaluating the similarity between the chiral template and the workpiece category. The weight coefficients ai and gs are dynamically adjusted according to the importance and influence of each indicator in the attitude control process to ensure the accuracy and rationality of the similarity calculation.

[0035] In this embodiment, the first similarity conversion coefficient d1 and the second similarity conversion coefficient d2 are used to convert the sum of the differences in the evaluation values ​​of the feature evaluation indicators and the sum of the differences in the evaluation values ​​of the demand evaluation indicators into a value with the same dimension as the similarity. The smaller the sum of the differences in the evaluation values, the greater the corresponding similarity, and vice versa.

[0036] In this embodiment, when dynamically adjusting the parameter template, the system linearly or non-linearly scales the key parameters (such as clamping force and rotation angle range) in the parameter template according to the actual size information of the new workpiece, so as to ensure that the parameter configuration meets the process requirements and avoids excessive constraints.

[0037] In some embodiments of this application, several candidate poses for generating a chiral template are included: Several first candidate poses of the chiral template are generated based on preset coarse-grained sampling parameters; For each first candidate pose, contour analysis is performed to extract contour features, including contour shape, contour size, and contour curvature. Compare the degree of difference in contour features of different first candidate poses, sort the contour features according to the degree of difference, and set several feature matching points for the corresponding contour features based on the sorting results. Each first candidate pose is projected onto the 3D point cloud image of the new workpiece, and the matching error between each contour feature of the first candidate pose and several feature matching points of the corresponding contour feature in the 3D point cloud image is calculated. The comprehensive matching error of the corresponding contour feature is generated based on the matching error of all feature matching points of each contour feature, and the effective coefficient of the corresponding first candidate pose is generated by combining the weight coefficient of the corresponding contour feature. Eliminate the first candidate pose whose effective coefficient is less than the preset first effective coefficient, and determine the effective region based on the remaining first candidate poses; Fine-grained sampling is performed on the effective region, and the candidate poses of the chiral template are determined based on the sampling results and the remaining first candidate poses.

[0038] In this embodiment, the preset coarse-grained sampling parameters include coarse step size, sampling interval, sampling angle range, etc. By setting reasonable sampling parameters, it is possible to reduce unnecessary computation while ensuring the diversity of candidate postures. The contour shape is used to describe the overall shape of the initial candidate posture, the contour size is used to measure the size of the initial candidate posture, and the contour curvature is used to reflect the degree of curvature of the initial candidate posture contour.

[0039] In this embodiment, a preset coarse-grained sampling parameter is used to quickly generate an initial candidate pose set covering the possible pose range of the chiral template, thereby improving computational efficiency by reducing the sampling density. Fine-grained sampling parameters, on the other hand, perform high-density sampling on the effective region to ensure the accuracy and diversity of candidate poses.

[0040] In this embodiment, the matching error is calculated based on the distance error and angle error of all feature matching points of the same contour feature.

[0041] In this embodiment, the calculation of the effective coefficient takes into account the matching error and sorting weight of each contour feature, effectively eliminating the initial candidate postures that deviate from the actual workpiece posture, and the final generated candidate postures provide high-quality basic data for subsequent posture screening.

[0042] In this embodiment, the greater the difference in contour features, the more different the coarse-grained sampling areas of the initial candidate poses are, laying the foundation for determining the effective area. The effective area is the area where the first candidate pose and the new workpiece are highly matched.

[0043] In this embodiment, the weight coefficient of the contour feature is set according to the sorting result. The higher the ranking (i.e. the greater the difference), the larger the weight coefficient, and vice versa.

[0044] In this embodiment, when the comprehensive matching error of the contour features is larger and the weight coefficient is larger, the corresponding effective coefficient is smaller, and vice versa. Furthermore, the preset first effective coefficient is smaller than the preset second effective coefficient.

[0045] In some embodiments of this application, fine-grained sampling is performed on the effective region, and candidate poses of the chiral template are determined based on the sampling results and the remaining first candidate poses, including: Based on the functional characteristics and task requirements of the new workpiece, the feature direction is determined. It is then determined whether the effective area covers the feature direction. If not, several first candidate postures are generated around the feature direction and the effective coefficient is calculated until the effective area covers the feature direction. If so, the effective area is sampled based on the preset fine-grained sampling parameters to obtain several second candidate postures. A multi-dimensional feature matching is performed on several second candidate poses, and an effective coefficient for each second candidate pose is generated based on the matching results. The multi-dimensional features include local point features, texture features, and topological structure features. Eliminate second candidate postures whose effective coefficient is less than the preset second effective coefficient; The remaining first candidate poses and the remaining second candidate poses are compared. Duplicate poses are eliminated based on the comparison results. The remaining first candidate poses and the remaining second candidate poses are set as the final candidate poses of the chiral template.

[0046] In this embodiment, the final candidate pose refers to a set of candidate solutions that are initially selected and may conform to the spatial pose (position + attitude) of the target workpiece in the real scene.

[0047] In this embodiment, the characteristic direction refers to the key orientation or motion trend direction that the new workpiece needs to maintain when performing a specific task. This direction is usually closely related to the workpiece's functional realization, assembly relationship, or motion constraints. For example, for a chiral workpiece that needs to be inserted into a specific slot, its characteristic direction may be the axial direction of the slot; for a workpiece that needs to rotate in coordination with other components, its characteristic direction may be the direction of the rotation axis. By clearly defining the characteristic direction, it can be ensured that the generated candidate posture not only matches the new workpiece geometrically, but also meets the task requirements in terms of functional realization.

[0048] In this embodiment, multi-dimensional feature matching refers to local point feature matching, texture feature matching, and topological structure feature matching. Specifically, it includes: pre-setting several local points of the second candidate feature; comparing the coordinates of each local point of the second candidate pose; calculating the Euclidean distance between it and the corresponding point in the 3D point cloud of the new workpiece; and calculating the average distance error of all local points as the local point feature matching error; extracting the texture features of the surface of the second candidate pose using a gray-level co-occurrence matrix; performing correlation analysis with the surface texture of the new workpiece; and generating a texture similarity index as the texture feature matching error; identifying features such as holes and protrusions of the second candidate pose using a topological structure analysis algorithm; comparing them with the topological structure of the new workpiece; and calculating the feature overlap rate as the topological structure feature matching error. The above three feature matching errors are weighted and summed according to preset weight coefficients to obtain the multi-dimensional feature comprehensive matching error of each second candidate pose, thereby generating the effective coefficient of each second candidate pose.

[0049] In this embodiment, by generating candidate poses of the chiral template, a precise data foundation is provided for subsequent pose selection and optimization, ensuring that the adaptability of the chiral template to the new workpiece can be evaluated from multiple dimensions and angles, and greatly improving the accuracy and robustness of pose estimation.

[0050] In some embodiments of this application, several candidate poses are screened based on key points and pose configuration rules to determine the optimal pose, including: Determine the task requirements for the new workpiece; Based on task requirements and user interaction information, graphical annotation and language rule conversion are performed to obtain several key points and several posture configuration rules. Calculate the correlation coefficient between each attitude configuration rule and several key points. Set the key points with correlation coefficients greater than the preset correlation coefficient threshold as the associated key points of the attitude configuration rule, and generate the standard constraint conditions for each associated key point to satisfy the attitude configuration rule. The associated key point sequence for each attitude configuration rule is generated sequentially. The order of the associated key point sequence is set according to the association coefficient, and each associated key point is mapped to a corresponding weight coefficient and standard constraint conditions. Each pose matching rule is treated as an optimization objective, and the weight coefficient of each optimization objective is set according to the task requirements; For each candidate pose, perform key point matching analysis to determine whether the candidate pose satisfies the standard constraint conditions of each associated key point in the associated key point sequence of each optimization objective. Based on the judgment results, determine the first number of associated key points that satisfy the standard constraint conditions of the same optimization objective and the second number of associated key points that do not satisfy the standard constraint conditions of the same optimization objective. The optimization coefficient of the corresponding candidate pose is calculated based on the first and second quantities of the same candidate pose for all optimization objectives. Several candidate poses are sorted according to the optimization coefficient, and the candidate pose ranked first is set as the optimal pose.

[0051] In this embodiment, the task requirements cover specific requirements such as the processing type, assembly sequence, and quality inspection standards of the new workpiece. The user interaction information includes supplementary information such as the operator's experience feedback, on-site environmental limitations, and special process requirements. The task requirements and user interaction information are transformed into intuitive geometric symbols or text annotations through graphical annotation, and then parsed into key point and posture configuration rules that can be processed by a computer through language rule conversion.

[0052] In this embodiment, the posture configuration rules are derived from user-defined rules and the requirements of downstream tasks (such as gripping and assembly), such as "the final posture must make the gripping face the robotic arm" or "the threaded hole of the screw must be aligned with the target hole position".

[0053] In this embodiment, the correlation coefficient is calculated by analyzing the degree of matching between the geometric constraints, motion restrictions and process requirements involved in the attitude configuration rules and the spatial location, functional attributes and process correlation of key points. The preset correlation coefficient threshold is dynamically adjusted according to the actual process accuracy requirements and system calculation efficiency.

[0054] In this embodiment, the standard constraints include spatial location constraints (such as key points must be located within the support area of ​​the candidate posture), motion direction constraints (such as the motion direction of key points must be consistent with the direction specified in the posture configuration rules), and process parameter constraints (such as the processing parameters at key points must meet the threshold range set in the posture configuration rules).

[0055] In this embodiment, the optimization coefficient reflects the degree to which the candidate poses comprehensively satisfy all optimization objectives. When sorting, the candidate poses with high optimization coefficients and strong stability are preferentially selected as the optimal solution.

[0056] In this embodiment, by pre-determining candidate postures and then accurately screening them based on key points and posture configuration rules, the intelligence level of chiral workpiece posture control is effectively improved, and the adaptability and robustness in multiple application scenarios are significantly enhanced. In actual operation, the system can automatically parse the task requirements and user interaction information of new workpieces, quickly generate key points and posture configuration rules that meet process requirements, and then evaluate candidate postures from multiple dimensions by calculating correlation coefficients and standard constraints. In this process, the weight coefficient setting of the optimization target ensures flexible adjustment under different process requirements, while the calculation of the optimization coefficient directly reflects the comprehensive performance of the candidate posture, thereby determining the optimal posture and greatly improving control efficiency.

[0057] In some embodiments of this application, the optimization coefficient of the corresponding candidate pose is calculated based on a first number and a second number of the same candidate pose for all optimization objectives, including: The formula for calculating the preference coefficient is: ; Where Y is the optimization coefficient, and w is the number of optimization targets. This represents the first number of associated key points that satisfy the v-th optimization objective for the corresponding candidate pose. Let y1 be the second number of associated key points corresponding to candidate poses that do not meet the v-th optimization objective, and y2 be the first preferred transformation coefficient and y2 be the second preferred transformation coefficient. The weight coefficient of the c-th associated key point in the v-th optimization objective is determined by... is the weight coefficient for the v-th optimization objective.

[0058] In this embodiment, the first preferred conversion coefficient refers to converting the quantity ratio into a value with the same dimension as the preferred coefficient. When the quantity ratio is larger, the corresponding preferred coefficient is larger, and vice versa. The second preferred conversion coefficient refers to converting the weight coefficient and value into a value with the same dimension as the preferred coefficient. When the weight coefficient and value are larger, the corresponding preferred coefficient is larger, and vice versa.

[0059] In this embodiment, by calculating the optimization coefficient of the candidate posture, it is possible to achieve a quantitative evaluation of the candidate posture's satisfaction with all optimization objectives. This enables the evaluation of a large number of candidate postures to be processed in a short time, providing a strong guarantee for the real-time performance and accuracy of chiral workpiece posture control.

[0060] In some embodiments of this application, workpiece posture data is acquired in real time, and combined with the optimal posture, dual-channel feature extraction results, and parameter configuration information, it is determined whether to generate a posture control command, including: Determine the real-time attitude based on the workpiece attitude data; Calculate the degree of positional deviation, angular deviation, and motion trend deviation between the real-time attitude and the optimal attitude, and quantify them into a first deviation coefficient, a second deviation coefficient, and a third deviation coefficient; Based on the dual-channel feature extraction results, the geometric feature deviation and chiral feature deviation of the workpiece surface are obtained respectively, and quantified into the fourth deviation coefficient and the fifth deviation coefficient. Determine the parameter value range of the chiral template in the parameter configuration information, and determine whether the real-time attitude exceeds the parameter value range. If so, quantize to obtain the sixth deviation coefficient. The above deviation coefficients and their corresponding weighting coefficients are weighted and summed to obtain the comprehensive deviation coefficient. Whether to generate attitude control commands is determined based on the comprehensive deviation coefficient.

[0061] In this embodiment, the greater the degree of positional deviation, angular deviation, and motion trend deviation, the greater the quantified first deviation coefficient, second deviation coefficient, and third deviation coefficient.

[0062] In this embodiment, the dual-channel feature extraction results are obtained by independently analyzing the workpiece surface geometry and chiral structure characteristics. The geometric feature deviations include morphological deviations such as surface flatness and contour symmetry, while the chiral feature deviations include topological attribute deviations such as left and right chiral structure matching degree and mirror symmetry. The quantification of the fourth and fifth deviation coefficients is achieved by comparing the deviation degree between real-time posture features and template features. When a local geometric deformation exceeds the preset tolerance or an asymmetric shift occurs in the chiral structure, the corresponding deviation coefficient will increase significantly.

[0063] In this embodiment, when each parameter of the real-time attitude is not in the corresponding parameter value range, the sixth deviation coefficient is calculated based on the parameter weight and the parameter difference when the parameter is not in the parameter value range. The larger the parameter difference and the larger the parameter weight, the larger the sixth deviation coefficient.

[0064] In this embodiment, the comprehensive deviation coefficient is obtained by weighting and summing the first to sixth deviation coefficients according to preset weights. This coefficient comprehensively reflects the degree of deviation between the real-time attitude and the optimal attitude in multiple dimensions such as spatial position, angle, motion trend, geometric features and chiral features.

[0065] In this embodiment, by acquiring workpiece posture data in real time and combining it with the optimal posture, dual-channel feature extraction results, and parameter configuration information for comprehensive judgment, precise control and dynamic adjustment of the chiral workpiece posture are achieved, effectively improving control accuracy and control efficiency.

[0066] In some embodiments of this application, determining whether to generate an attitude control command based on a comprehensive deviation coefficient includes: Pre-set the overall deviation coefficient threshold; If the overall deviation coefficient is less than the overall deviation coefficient threshold, an attitude control command is generated. If the overall deviation coefficient is not less than the overall deviation coefficient threshold, no attitude control command will be generated.

[0067] In this embodiment, the comprehensive deviation coefficient threshold is based on the maximum comprehensive deviation coefficient that meets both the actual process accuracy requirements and the task requirements. In this embodiment, when the overall deviation coefficient exceeds the threshold, the system determines that the current posture deviates from the optimal state and immediately triggers a control command; otherwise, no control command is triggered. The posture control command includes, but is not limited to, parameters such as the workpiece's rotation angle and translation distance, ensuring that the workpiece quickly and accurately reaches the optimal posture. Simultaneously, the system continuously monitors the workpiece's posture data, forming a closed-loop control until the overall deviation coefficient stabilizes below the threshold, thereby ensuring that the chiral workpiece maintains high-precision and high-stability posture control in complex and ever-changing scenarios.

[0068] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and substitutions can be made without departing from the technical principles of this application, and these improvements and substitutions should also be considered within the scope of protection of this application.

Claims

1. A method for controlling the posture of a chiral workpiece supporting multiple application scenarios, characterized in that, include: Receive the CAD model of the new workpiece and analyze its geometric features. Determine the chiral symmetry information based on the analysis results and generate the corresponding mirror model as a chiral template. Parameter templates for different workpiece categories are pre-built, the workpiece category of the chiral template is determined, and dynamic adaptation is performed in combination with the corresponding parameter template to obtain the parameter configuration information of the chiral template. Generate several candidate poses for the chiral template, mark key points according to task requirements and generate pose configuration rules, filter several candidate poses according to key points and pose configuration rules, determine the optimal pose and issue pose control commands. The system acquires workpiece posture data in real time, and combines the optimal posture, dual-channel feature extraction results, and parameter configuration information to determine whether to generate a posture adjustment command.

2. The chiral workpiece posture control method supporting multi-scenario applications as described in claim 1, characterized in that, Receive the CAD model of the new workpiece and analyze its geometric features. Based on the analysis results, determine the chiral symmetry information and generate a corresponding mirror model as a chiral template, including: The CAD model of the new workpiece is received and standardized, including coordinate normalization, mesh simplification and noise removal. Extract the geometric features of the processed CAD model, including curvature features, normal vector features, feature lines, and topological features; Analyze the geometric features and determine the candidate axes of symmetry and candidate planes of symmetry based on the analysis results; Calculate the degree of symmetry of candidate axes of symmetry and candidate planes of symmetry; Candidate symmetry axes and candidate symmetry planes with a symmetry degree greater than a preset symmetry degree threshold are set as chiral symmetry information, and a mirror model is generated. Calculate the matching degree between the mirror model and the CAD model; If the matching degree is greater than the preset matching degree threshold, the mirror model is set as a chiral template. If the matching degree is not greater than the preset matching degree threshold, the mirror model is corrected until the matching degree is greater than the preset matching degree threshold, and the corresponding mirror model is set as a chiral template.

3. The chiral workpiece posture control method supporting multi-scenario applications as described in claim 2, characterized in that, Pre-build parameter templates for different workpiece categories, including: Feature extraction is performed on historical workpiece data to obtain several workpiece features; Multiple workpiece categories are generated based on all workpiece features, and a parameter feature reference library is constructed for each workpiece category; Several demand evaluation indicators are pre-defined; The parameter feature reference library for each workpiece category is evaluated based on several demand evaluation indicators to obtain the demand evaluation value for each demand evaluation indicator. The demand evaluation indicators for the same workpiece category are sorted according to the demand evaluation value to obtain a demand evaluation indicator sequence; Based on the ranking results in the demand evaluation index sequence and the corresponding demand evaluation values, key parameters and corresponding parameter value ranges are set for each workpiece category. A parameter template for each workpiece category is constructed based on the key parameters and corresponding parameter value ranges for each workpiece category.

4. The chiral workpiece posture control method supporting multi-scenario applications as described in claim 3, characterized in that, The workpiece category of the chiral template is determined and dynamically adapted in conjunction with the corresponding parameter template to obtain the parameter configuration information of the chiral template, including: Extract real-time workpiece features from chiral templates; The real-time workpiece features are compared with several workpiece features of each workpiece category to obtain several similarities between the chiral template and different workpiece categories. The formula for calculating the similarity is: ； Where D represents the similarity, and n1 represents the number of feature evaluation indicators. t2i is the reference evaluation value of the i-th feature evaluation index generated for real-time workpiece features, t2i is the historical evaluation value of the i-th feature evaluation index generated for workpiece features, ai is the weight coefficient of the i-th feature evaluation index, and n2 is the number of demand evaluation indicators. The reference demand evaluation value for the s-th demand evaluation index generated for real-time workpiece features. The demand evaluation value of the s-th demand evaluation index generated for the workpiece features, gs is the weight coefficient of the s-th demand evaluation index, d1 is the first similarity conversion coefficient, and d2 is the second similarity conversion coefficient; All workpiece categories are sorted according to similarity, and the workpiece category with the highest similarity is set as the workpiece category of the chiral template. Based on the workpiece category and the size information of the new workpiece, the parameter template of the corresponding workpiece category is loaded and dynamically adjusted to generate the parameter configuration information of the chiral template.

5. The chiral workpiece posture control method supporting multi-scenario applications as described in claim 4, characterized in that, Several candidate poses for generating the chiral template are included: Several first candidate poses of the chiral template are generated based on preset coarse-grained sampling parameters; For each first candidate pose, contour analysis is performed to extract contour features, including contour shape, contour size, and contour curvature. Compare the degree of difference in contour features of different first candidate poses, sort the contour features according to the degree of difference, and set several feature matching points for the corresponding contour features based on the sorting results. Each first candidate pose is projected onto the 3D point cloud image of the new workpiece, and the matching error between each contour feature of the first candidate pose and several feature matching points of the corresponding contour feature in the 3D point cloud image is calculated. The comprehensive matching error of the corresponding contour feature is generated based on the matching error of all feature matching points of each contour feature, and the effective coefficient of the corresponding first candidate pose is generated by combining the weight coefficient of the corresponding contour feature. Eliminate the first candidate pose whose effective coefficient is less than the preset first effective coefficient, and determine the effective region based on the remaining first candidate poses; Fine-grained sampling is performed on the effective region, and the candidate poses of the chiral template are determined based on the sampling results and the remaining first candidate poses.

6. The chiral workpiece posture control method supporting multi-scenario applications as described in claim 5, characterized in that, Fine-grained sampling is performed on the effective region, and candidate poses of the chiral template are determined based on the sampling results and the remaining first candidate poses, including: Based on the functional characteristics and task requirements of the new workpiece, the feature direction is determined. It is then determined whether the effective area covers the feature direction. If not, several first candidate postures are generated around the feature direction and the effective coefficient is calculated until the effective area covers the feature direction. If so, the effective area is sampled based on the preset fine-grained sampling parameters to obtain several second candidate postures. A multi-dimensional feature matching is performed on several second candidate poses, and an effective coefficient for each second candidate pose is generated based on the matching results. The multi-dimensional features include local point features, texture features, and topological structure features. Eliminate second candidate postures whose effective coefficient is less than the preset second effective coefficient; The remaining first candidate poses and the remaining second candidate poses are compared. Duplicate poses are eliminated based on the comparison results. The remaining first candidate poses and the remaining second candidate poses are set as the final candidate poses of the chiral template.

7. The chiral workpiece posture control method supporting multi-scenario applications as described in claim 6, characterized in that, Based on key points and attitude configuration rules, several candidate attitudes are screened to determine the optimal attitude, including: Determine the task requirements for the new workpiece; Based on task requirements and user interaction information, graphical annotation and language rule conversion are performed to obtain several key points and several posture configuration rules. Calculate the correlation coefficient between each attitude configuration rule and several key points. Set the key points with correlation coefficients greater than the preset correlation coefficient threshold as the associated key points of the attitude configuration rule, and generate the standard constraint conditions for each associated key point to satisfy the attitude configuration rule. The associated key point sequence for each attitude configuration rule is generated sequentially. The order of the associated key point sequence is set according to the association coefficient, and each associated key point is mapped to a corresponding weight coefficient and standard constraint conditions. Each pose matching rule is treated as an optimization objective, and the weight coefficient of each optimization objective is set according to the task requirements; For each candidate pose, perform key point matching analysis to determine whether the candidate pose satisfies the standard constraint conditions of each associated key point in the associated key point sequence of each optimization objective. Based on the judgment results, determine the first number of associated key points that satisfy the standard constraint conditions of the same optimization objective and the second number of associated key points that do not satisfy the standard constraint conditions of the same optimization objective. The optimization coefficient of the corresponding candidate pose is calculated based on the first and second quantities of the same candidate pose for all optimization objectives. Several candidate poses are sorted according to the optimization coefficient, and the candidate pose ranked first is set as the optimal pose.

8. The chiral workpiece posture control method supporting multi-scenario applications as described in claim 7, characterized in that, The optimization coefficient of the corresponding candidate pose is calculated based on the first and second quantities of the same candidate pose for all optimization objectives, including: The formula for calculating the preference coefficient is: ； Where Y is the optimization coefficient, and w is the number of optimization targets. This represents the first number of associated key points that satisfy the v-th optimization objective for the corresponding candidate pose. Let y1 be the second number of associated key points corresponding to candidate poses that do not meet the v-th optimization objective, and y2 be the first preferred transformation coefficient and y2 be the second preferred transformation coefficient. The weight coefficient of the c-th associated key point in the v-th optimization objective is determined by... is the weight coefficient for the v-th optimization objective.

9. The chiral workpiece posture control method supporting multi-scenario applications as described in claim 8, characterized in that, Real-time acquisition of workpiece posture data, combined with optimal posture, dual-channel feature extraction results, and parameter configuration information, determines whether to generate posture control commands, including: Determine the real-time attitude based on the workpiece attitude data; Calculate the degree of positional deviation, angular deviation, and motion trend deviation between the real-time attitude and the optimal attitude, and quantify them into a first deviation coefficient, a second deviation coefficient, and a third deviation coefficient; Based on the dual-channel feature extraction results, the geometric feature deviation and chiral feature deviation of the workpiece surface are obtained respectively, and quantified into the fourth deviation coefficient and the fifth deviation coefficient. Determine the parameter value range of the chiral template in the parameter configuration information, and determine whether the real-time attitude exceeds the parameter value range. If so, quantize to obtain the sixth deviation coefficient. The above deviation coefficients and their corresponding weighting coefficients are weighted and summed to obtain the comprehensive deviation coefficient. Whether to generate attitude control commands is determined based on the comprehensive deviation coefficient.

10. The chiral workpiece posture control method supporting multi-scenario applications as described in claim 9, characterized in that, Whether to generate attitude control commands is determined based on the comprehensive deviation coefficient, including: Pre-set the overall deviation coefficient threshold; If the overall deviation coefficient is less than the overall deviation coefficient threshold, an attitude control command is generated. If the overall deviation coefficient is not less than the overall deviation coefficient threshold, no attitude control command will be generated.

Images